Downscaling of species distribution models: 
a hierarchical approach

Abstract

Summary1. Reliable methods to downscale species distributions from coarse to fine grain (equivalent to resolution or support) hold great potential benefit for ecology and conservation. Existing methods have been based on partially unrealistic assumptions and yield mixed results. 2. Here, we introduce a novel and simple approach for downscaling species distribution models based on a hierarchical Bayesian modelling (HBM) framework. Our approach treats putative (unknown) fine-grain presences/absences as latent variables, which are modelled as a function of observed fine-grain environmental variables and constrained by observed coarse-grain presences/absences using logistic regression. The aim is to produce downscaled fine-grain probabilities of species occurrence that (1) closely resemble the probabilities produced by a logistic model parameterized with the observed fine-grain data (the 'reference model') and (2) are improvements over conventional downscaling methods. We additionally test how fine-grain occupancy based on power-law scale-area relationships modifies the downscaling results. We test our approach on 127 bird species from the San Diego breeding atlas data surveyed at 5 km grain. 3. The HBM approach provides unbiased fine-grain probabilities of occurrence whilst the conventional methods (direct approach, point sampling) consistently over-predict occurrence probabilities. Incorporation of the downscaled occupancy further improves reliability of the models, but only in cases when the fine-grain occupancy is estimated accurately. 4. Summing predictions across grid cells and species, HBMs provide better estimates of fine-grain species richness than conventional methods. They also provide better estimates of fine-grain occupancy (prevalence). 5. The presented HBM-based downscaling approach offers improved predictions of fine-grain presence and absence compared with existing methods. The combination of the Bayesian approach with key macroecological relationships (specifically, the scale-area relationship) offers a promising general basis for downscaling distributions that may be extended, for example, using generalized linear or additive models. These approaches enable integrative predictions of spatial biodiversity patterns at fine grains.

Monitoring plant functional diversity from space

“…The increasing volume of online species occurrence data is a fourth synergistic development that supports the predictive modeling and mapping of species' and plant community distributions 57 . Combined with trait and phylogenetic data, and potentially other ecological information (such as typical stand density), hierarchical statistical models and downscaling techniques 58,59 may, with some uncertainty, allow the pinpointing of particular species and the make-up of communities. We hypothesize that such predictions will generally be much more effective at coarser levels of biological organization, such as higher-level clades or other wellcharacterized species groups that can be associated with the aggregate functions of the spectral signal of a pixel.…”

Spatial Downscaling of Alien Species Presences Using Machine Learning

“…Assumptions made by direct and iterative approaches (Keil et al, 2013) are often criticized for speculating similar species association with environmental variables across scales (Menke et al, 2009). On the other hand, point sampling and clustering approaches make assumptions regarding habitat suitability within the coarse grid (Keil et al, 2013). In their review, Trzaska and Schnarr (2014) distinguish downscaling methods between the relatively straight-forward but normality-limited linear methods (e.g., delta method, simple and multiple linear regression, canonical-correlation analysis, etc.)…”

“…Although, it has extensively been used in other scientific disciplines (Trzaska and Schnarr, 2014), downscaling is not a trivial process. Keil et al (2013) lists four strictly computational approaches that have been used to predict fine-gridded species presence based on a coarse grid. Assumptions made by direct and iterative approaches (Keil et al, 2013) are often criticized for speculating similar species association with environmental variables across scales (Menke et al, 2009).…”

“…Keil et al (2013) lists four strictly computational approaches that have been used to predict fine-gridded species presence based on a coarse grid. Assumptions made by direct and iterative approaches (Keil et al, 2013) are often criticized for speculating similar species association with environmental variables across scales (Menke et al, 2009). On the other hand, point sampling and clustering approaches make assumptions regarding habitat suitability within the coarse grid (Keil et al, 2013).…”

Erratum: Monitoring plant functional diversity from space

“…Combined with trait and phylogenetic data, and potentially other ecological information (such as typical stand density), hierarchical statistical models and downscaling techniques 58,59 may, with some uncertainty, allow the pinpointing of particular species and the make-up of communities. We hypothesize that such predictions will generally be much more effective at coarser levels of biological organization, such as higher-level clades or other wellcharacterized species groups that can be associated with the aggregate functions of the spectral signal of a pixel.…”